Montage Technology Begins Industry-First Trial Production of CXL 3.2 MXC Chip to Advance AI and Data Center Memory Innovation
HENGQIN, GUANGDONG, China — Montage Technology has announced the successful industry-first trial production of its Compute Express Link (CXL®) 3.2 Memory eXpander Controller (MXC) chip, marking a significant milestone in the evolution of memory expansion technologies for artificial intelligence (AI), cloud computing, and modern data center infrastructure. The breakthrough highlights the company’s continued investment in next-generation memory architecture designed to address the rapidly increasing demand for scalable, high-performance computing.
As AI applications, large language models (LLMs), high-performance computing (HPC), and cloud-native workloads continue to push the limits of traditional server architectures, memory capacity and bandwidth have become critical performance bottlenecks. Montage Technology’s new CXL 3.2 MXC chip is engineered to overcome these limitations by enabling flexible memory expansion, intelligent resource pooling, and improved system scalability while maintaining high performance and reliability.
Addressing the Growing Demand for Memory Expansion
The exponential growth of AI-driven applications has dramatically increased the need for larger memory capacities capable of supporting complex data processing and machine learning workloads. Traditional server memory architectures often limit scalability because memory resources remain tightly coupled to individual processors.
Montage Technology’s CXL 3.2 MXC addresses this challenge by utilizing the Compute Express Link (CXL) standard, an open industry interconnect designed to enable high-speed communication between processors, memory, and accelerators. By separating memory resources from compute resources, CXL allows organizations to build more flexible and efficient infrastructure capable of dynamically allocating memory where it is needed most.
The company’s latest MXC chip is designed to help cloud service providers, enterprise customers, and hyperscale data centers improve resource utilization while reducing infrastructure costs associated with memory overprovisioning.
Advanced CXL 3.2 Architecture
The newly introduced MXC controller complies fully with the CXL Type 3 specification, supporting both CXL.mem and CXL.io protocols. Built on the latest PCI Express (PCIe®) 6.x and CXL 3.2 standards, the chip delivers impressive data transfer speeds of up to 64 gigatransfers per second (GT/s), providing the bandwidth necessary for next-generation computing platforms.
The controller integrates dual DDR5 memory controllers capable of supporting DDR5 memory operating at speeds of up to 8000 MT/s. This enables the MXC chip to translate host-generated CXL memory requests into DDR commands in real time, allowing seamless communication between server processors and backend memory devices.
By enabling efficient data exchange and removing traditional server memory limitations, the chip significantly expands available system memory while maintaining low latency and high throughput. This capability is particularly valuable for AI training, inference, big data analytics, and memory-intensive enterprise workloads.
Designed for Flexible Memory Architectures
Beyond its high-speed performance capabilities, the CXL 3.2 MXC incorporates numerous advanced hardware components that support deployment across a wide range of enterprise environments.
The chip integrates:
- PCIe 6.x physical layer (PHY)
- DDR5 PHY technology
- Dual RISC-V processor subsystems
- Comprehensive management and monitoring interfaces
These integrated features enable the MXC controller to support multiple CXL memory expansion form factors, including PCIe Add-In Cards (AICs) and Enterprise and Data Center Standard Form Factor (EDSFF) modules.
This flexibility allows system designers to develop innovative memory architectures supporting use cases such as:
- Memory expansion
- Memory pooling
- Tiered memory architectures
- Shared memory infrastructure
- High-capacity AI servers
- Cloud-scale resource allocation
Such capabilities are becoming increasingly important as organizations seek to maximize server efficiency while supporting increasingly demanding AI workloads.
Accelerating CXL Commercialization
Stephen Tai, President of Montage Technology, described the trial production milestone as a major achievement in the company’s long-term CXL strategy.
According to Tai, the successful trial production of the CXL 3.2 MXC represents an important step toward commercializing advanced CXL technologies and expanding their adoption across AI infrastructure and modern data centers.
He noted that memory demand continues to accelerate rapidly in the AI era, requiring innovative technologies capable of delivering higher bandwidth, greater memory capacity, and improved reliability. Montage Technology’s latest solution has been designed to provide customers with scalable memory expansion capabilities while accelerating widespread deployment of CXL-enabled infrastructure.
The company believes that widespread adoption of CXL technologies will fundamentally reshape server memory architecture by enabling more efficient utilization of computing resources across enterprise and cloud environments.
Comprehensive Software Development Support
Recognizing that successful deployment extends beyond hardware innovation, Montage Technology also provides customers with a complete Software Development Kit (SDK) alongside comprehensive analysis, debugging, and testing tools.
These software resources are intended to simplify product development, reduce integration complexity, accelerate compatibility testing, and shorten the path toward commercial deployment.
By offering both hardware and software support, Montage Technology enables system manufacturers and solution providers to bring CXL-enabled products to market more quickly while ensuring interoperability across multiple computing platforms.
Collaboration with Industry Leaders
Montage Technology revealed that its CXL 3.2 MXC has already been integrated into next-generation CXL memory products developed by leading global memory manufacturers, including Samsung and SK hynix.
Initial validation of these products has been successfully completed, demonstrating the controller’s readiness for commercial deployment and its compatibility with advanced memory modules designed for future AI infrastructure.
The solution has also been engineered to support leading server processor platforms, including Intel Xeon and AMD EPYC, ensuring broad interoperability across enterprise computing environments.
This ecosystem approach enables customers to deploy CXL-enabled infrastructure with confidence while leveraging existing processor and memory technologies from industry-leading vendors.
Industry Leaders Highlight the Importance of CXL
Executives from several leading technology companies welcomed the advancement and emphasized the growing importance of CXL-enabled memory solutions.
Samsung Electronics highlighted that combining CXL 3.2 MXC technology with DDR5 memory provides data center operators with an effective balance between memory capacity and system performance. The company noted that the collaboration validates CXL’s ability to improve resource utilization while expanding system memory for AI infrastructure.
SK hynix emphasized that rapid adoption of generative AI and large language models has made memory one of the most critical components affecting system performance. The company believes Montage Technology’s MXC controller demonstrates strong potential for enabling high-capacity memory expansion and efficient resource sharing, while expressing interest in expanding collaboration on future ecosystem development.
Intel also praised the continued maturity of the CXL ecosystem. Company representatives noted that AI workloads, next-generation cloud computing, and evolving server economics are driving demand for more flexible memory architectures. Montage Technology’s CXL 3.2 MXC was recognized for demonstrating robust protocol support, strong interoperability, and performance capabilities that can unlock advanced memory expansion on future Intel Xeon platforms.
AMD similarly emphasized that AI and enterprise data center workloads continue reshaping modern computing architecture. According to the company, scalable memory infrastructure has become increasingly important for improving performance and enabling future system designs. AMD reaffirmed its commitment to collaborating with Montage Technology to advance CXL technologies capable of supporting more efficient and flexible memory ecosystems.
Showcasing Innovation at Industry Events
To demonstrate its latest innovation, Montage Technology will showcase the CXL 3.2 MXC chip at two major industry events in the United States: CXL DevCon and the Future of Memory and Storage (FMS) conference.
The exhibitions will provide customers, technology partners, hardware manufacturers, and industry experts with an opportunity to experience the new solution firsthand, discuss emerging trends in memory architecture, and explore opportunities for technical collaboration.
By participating in these leading technology conferences, Montage Technology aims to strengthen industry partnerships while promoting broader adoption of CXL technologies across global data center and AI ecosystems.
Building the Future of AI Memory Infrastructure
As artificial intelligence continues transforming industries worldwide, the importance of scalable, high-performance memory infrastructure has never been greater. Organizations deploying increasingly sophisticated AI models require innovative technologies capable of overcoming traditional memory limitations while maximizing infrastructure efficiency.
With the successful trial production of its CXL 3.2 Memory eXpander Controller, Montage Technology has taken an important step toward enabling the next generation of intelligent computing. By combining high-speed connectivity, advanced DDR5 support, broad ecosystem compatibility, and flexible deployment options, the company is helping shape a future where memory resources can be dynamically allocated, efficiently utilized, and seamlessly integrated across modern computing platforms.
The achievement reinforces Montage Technology’s commitment to advancing CXL innovation and delivering solutions that empower AI, cloud computing, and enterprise customers to build faster, smarter, and more scalable digital infrastructure for the years ahead.
Source link: https://www.businesswire.com/